Image Classification on ImageNet 1k (val) (Accuracy & ACEv2/Energy Efficiency)
73.59AccuracyPikeLPN-6x
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PikeLPN-6xUsed Precisions=16, 8, 42024.03 | 73.59 | 58.74 | 98.87 | 1.13 | 243.85 | 63.38 | |
| PokeBNN-1xUsed Precisions=32, 8, 4, 12024.03 | 73.4 | 68.56 | 6.16 | 93.83 | 298.44 | 40.48 | |
| PikeLPN-3xUsed Precisions=16, 8, 42024.03 | 71.95 | 33.7 | 98.52 | 1.48 | 139.59 | 52.66 | |
| MobileNetMode=8bit, Used Precisions=32, 82024.03 | 70.7 | 51.44 | 79.61 | 20.39 | 173.68 | 39.57 | |
| PokeBNN-0.75xUsed Precisions=32, 8, 4, 12024.03 | 70.5 | 50.61 | 5.11 | 94.88 | 218.51 | 40.48 | |
| ReActNetUsed Precisions=32, 12024.03 | 69.4 | 83.24 | 26.78 | 73.22 | 361.63 | 36.75 | |
| PikeLPN-2xUsed Precisions=16, 8, 42024.03 | 69.23 | 15.56 | 97.87 | 2.13 | 64.2 | 39.57 | |
| MeliusNet-42Used Precisions=32, 12024.03 | 69.2 | 215.71 | — | — | 901.82 | — | |
| PROFITUsed Precisions=32, 42024.03 | 69.05 | 20.91 | 47.51 | 52.49 | 82.7 | 39.57 | |
| PikeLPN-1xUsed Precisions=8, 42024.03 | 67.55 | 8.5 | 96.38 | 3.62 | 34.98 | 39.57 | |
| MeliusNet-29Used Precisions=32, 12024.03 | 65.8 | 158.21 | — | — | 656.81 | — | |
| Real-to-Binary NetUsed Precisions=32, 12024.03 | 65.4 | 186.85 | — | — | 762.24 | — | |
| PokeBNN-0.5xUsed Precisions=32, 8, 4, 12024.03 | 65.2 | 33.58 | 4.18 | 95.81 | 143.78 | 24.5 | |
| MobileNetMode=4W, 8A, Used Precisions=32, 8, 42024.03 | 65 | 33.8 | 68.96 | 31.04 | 118.54 | 39.57 | |
| MobileNetMode=8W, 4A, Used Precisions=32, 8, 42024.03 | 64 | 33.8 | 68.96 | 31.04 | 118.54 | 39.57 | |
| Bi-RealNet-34Used Precisions=32, 12024.03 | 62.2 | 168.11 | — | — | 691.47 | — | |
| Bi-RealNet-18Used Precisions=32, 12024.03 | 56.4 | 166.26 | — | — | 678.75 | — | |
| MobiNetUsed Precisions=32, 12024.03 | 54.4 | 12.64 | 13.17 | 86.83 | 50.66 | 28 | |
| XNOR-NetUsed Precisions=32, 12024.03 | 51.2 | 143.78 | — | — | 587.69 | — |